JP2024506605A - Systems and methods for improving soft tissue contrast, multiscale modeling and spectral CT - Google Patents

Abstract

Systems and methods are provided for improving soft tissue contrast, characterizing tissues, classifying phenotypes, stratifying risk, and performing multiscale modeling assisted by multi-energy or multi-contrast excitation and evaluation. . The systems and methods include assessing atherosclerotic plaque tissue in, for example, blood vessel walls and perivascular spaces using monophasic and multiphasic acquisition as well as broad and local spectral imaging.
[Selection diagram] Figure 4

Description

(Cross reference to related applications)
This application claims the benefit and priority of U.S. Provisional Patent Application No. 63/147,609, filed February 9, 2021, the entire contents of which are incorporated herein by reference in their entirety. incorporated herein and owned by the assignee of this application.

(Government license right)
This research was supported in part by the National Heart, Lung, and Blood Institute under contract number HL 126224. The United States Government may have certain rights in this invention.

The present invention relates to computer-assisted phenotyping (CAP) of disease, which involves applying computerized image analysis and/or data fusion algorithms to patient data. In particular, the present invention relates to quantitative imaging and analysis to elucidate the atherosclerotic disease process, including techniques to improve soft tissue contrast, multiscale modeling and/or spectral CT.

Atherosclerosis can be life-threatening, especially in the elderly, but also in relatively young people. Currently, there are limited methods for diagnosing atherosclerosis (e.g., the use of blood markers such as cholesterol levels) and/or determining the extent of luminal narrowing (stenosis), which may lead to optimal treatment. Unable to make legal decisions (eg, to perform or not perform surgery, or prescribe intensive drug therapy). For example, many vascular surgeries do not help patients, some patients require surgery but cannot undergo it, and others can be effectively treated with drugs but are not prescribed.

Current tools can analyze the vascular lumen, but do not accurately diagnose atherosclerosis because atherosclerosis is a disease of the blood vessel wall, not the blood and the passageway through which blood flows. may be insufficient. Misclassification of risk levels, failure to assess expected response to drug therapy, and/or failure to measure response to drug are likely to occur.

Radiological imaging can now be used as a non-invasive and safe method to identify the cause of disease. Current medical imaging tools include computed tomography (CT, including single-energy, multi-energy, or spectral CT), magnetic resonance imaging (MR, including MRA, DCE-MRI, or multi-contrast MRI), ultrasound (b-mode or intravascular US), and targeted contrast approaches using various imaging modalities.

Advances in imaging technology have made medical imaging an integral part of patient care. Imaging is valuable because it can provide spatially and temporally localized anatomical and/or functional information using non-invasive or minimally invasive methods. However, addressing increased resolution, such as exploiting patterns and/or signatures in data that are not normally easily assessed by the human eye, and managing large amounts of data to efficiently integrate it into clinical workflows, Technology is required. New high-resolution imaging techniques allow radiologists to ``drowse'' in the data without assistance. Thus, for example, to integrate quantitative imaging for individual patient management, decision support informatics tools are needed that make imaging capabilities more available within the workflow and/or reimbursement constraints realities of existing tools. It is preferable to provide a class.

Currently, imaging of atherosclerosis is routinely performed both invasively through catheter insertion and non-invasively through the use of ultrasound, CT, MR and nuclear medicine techniques. The most typical evaluation is luminal stenosis. Recent advances have been in the determination of myocardial flow reserve ratio.

Currently, one difficulty with imaging atherosclerosis is the lack of robustness of the methods used. For example, current methods typically have low contrast between the outer vessel wall and perivascular tissue, making it difficult to distinguish between the two. Current methods simply utilize an annular ring surrounding the lumen without specifically determining the boundaries of the outer wall. Tapering of blood vessels, branching vessels, and nearby tissue can also be problematic.

Another difficulty with current atherosclerosis imaging may be due to certain imaging devices that use limited excitation to acquire information from the tissue, use multi-contrast MR, or use multi-energy Despite using CT, there may be some non-specific reaction within the generated signal.

Advantages of the invention may include improving soft tissue contrast. Some advantages of the invention may include improved characterization of tissue. Some advantages of the invention may include classifying phenotypes. Some advantages of the invention may include risk stratification. Some advantages of the present invention may include performing multiscale modeling assisted by multi-energy or multi-contrast excitation and evaluation.

Some advantages of the present invention may include assessing atherosclerotic plaque tissue in vessel walls and perivascular spaces using single-relative multiphase acquisition and wide-spectrum CT. Some advantages of the present invention may include distributed tissue types as an overlay of locally organized tissue.

In one aspect, the invention includes a computerized method for improving soft tissue analysis. The method may further include the step of acquiring a plurality of radiographic images of the patient, each of the plurality of radiographic images being acquired using a different excitation. The method may further include selecting one treatment from a plurality of treatments to analyze the plurality of excitations based on an expected soft tissue type. The method may further include segmenting the processed plurality of excitations to display the soft tissue.

In some embodiments, the plurality of radiological images are computed tomography (CT) images and the different excitations are different x-ray energies. In some embodiments, the plurality of radiographic images are magnetic resonance (MR) images, and the different excitations are different radio frequency pulses.

In some embodiments, the plurality of radiographic images are ultrasound images and the different excitations are at different frequencies.

In some embodiments, the present invention provides the step of determining, by a computing device, a first tissue type in a region of interest based on the plurality of radiographic images of the patient, the step of determining a first tissue type in a region of interest. a type is represented by a grid of points across the region of interest; and determining, by the computing device, a second tissue type within the region of interest based on a plurality of radiographic images of the patient. , the second tissue type is a focal region within the region of interest, and at least some grid points of the first tissue type are dislocated from grid points of the second tissue type. and a matching step.

In some embodiments, the plurality of processes includes a digital subtraction process, a digital addition process, a multivariate statistical process, or an excitation selection process. In some embodiments, the digital subtraction process includes subtracting a first subset of the plurality of radiological images from one or more of the plurality of radiological images that are not within the subset. The digital addition process includes averaging the plurality of received radiographic images.

In some embodiments, the multivariate statistical processing includes combining the plurality of radiological images and removing inter-class dependencies using a multivariate statistical approach.

In some embodiments, the plurality of radiological images are CT images, and each of the plurality of CT images has a first A second x-ray source directs a second x-ray attenuator to an energy-integrating detector dimensioned to produce an image of photons to produce an image of a particular tissue target within the predetermined region. by generating a final CT image for a counting-type detector, by a processor, based on the image of the predetermined region of the patient and the image of a tissue target of the particular tissue within the predetermined image; .

In some embodiments, the excitation selection process includes selecting a particular radiographic image from the plurality of radiographic images based on the tissue type.

In other aspects, the invention includes a hybrid computed tomography (CT) scanner. The hybrid CT scanner may include a first x-ray source that directs a first x-ray attenuation to an energy integrating detector dimensioned to generate an image of a predetermined region of the patient. The hybrid CT scanner may further include a second x-ray source that directs a second x-ray attenuation to a photon counting detector that generates an image of a specific tissue target within the predetermined region. The hybrid CT scanner may further include a processor that generates a final computed tomography (CT) image based on the image of the predetermined region of the patient and the image of the particular tissue target within the predetermined image. .

In some embodiments, the energy integrating detector and the photon counting detector are arranged to acquire information from the same field of view. In some embodiments, the image of the predetermined region of the patient is a grayscale CT image. In some embodiments, the image of the particular tissue target within the predetermined image is a spectral CT image. In some embodiments, the energy integrating detector is positioned with a 90 degree offset between the photon counting detector and the second x-ray source with respect to the first x-ray source. be done

In some embodiments, the processor is configured to analyze tissue type, and the analysis includes one process from a plurality of processes to analyze the final CT image based on expected soft tissue types. and segmenting the final processed CT image to display the soft tissue.

In some embodiments, the photon counting detector includes multiple energy bins configured to image the particular tissue target.

In other aspects, the invention includes a computerized method for determining and displaying mixed tissue types. The method may include receiving, by a computing device, a radiological image of the patient. The method includes determining, by the computing device, a first tissue type within a region of interest based on the radiographic image of the patient, the first tissue type determining a point across the region of interest. It may include steps represented by a grid. The method includes determining, by the computing device, a second tissue type in the region of interest based on the radiographic image of the patient, the second tissue type in the region of interest. the at least some grid points of the first tissue type are coincident in position with grid points of the second tissue type.

In some embodiments, the plurality of radiographic images are computed tomography (CT) images, magnetic resonance (MR) images, or ultrasound images. In some embodiments, the first tissue type and the second tissue type are overlaid. In some embodiments, the grid of points has varying densities.

In some embodiments, the first tissue type is microcalcified tissue and the second tissue type is LNRC, dense calcification, or IPH.

In some embodiments, the method includes performing, by the computing device, multi-energy photon counting K-edge differential imaging on the CT image; and performing, by the computing device, on the CT image. denoising the spectral image using a regularization model; and improving the signal-to-noise ratio of calcium on the K-edge subtracted and denoised CT image by the computing device. , segmenting the improved calcium image by the computing device to represent one or both of localized dense calcifications and distributed microcalcifications.

Non-limiting examples of embodiments of the present disclosure will now be described with reference to the accompanying drawings of the specification listed after this paragraph. The dimensions of the features shown in the drawings have been chosen for convenience and clarity of presentation and are not necessarily drawn to scale.

The subject matter of the invention is particularly pointed out and distinctly claimed in the concluding portion of the specification. However, the structure and method of operation of the present invention, as well as objects, features and advantages thereof, may be better understood from the following detailed description, taken in conjunction with the accompanying drawings. Embodiments of the invention are shown by way of example, and not by way of limitation, in the accompanying drawings, in which like reference numerals indicate corresponding similar or similar elements.

2 is an exemplary radiation input and processed output diagram in accordance with some embodiments of the present invention; FIG. 2 is a diagram illustrating an example where atherosclerosis progresses with common drivers across different degrees of arterial beds based on evolutionary differences and local factors, in accordance with some embodiments of the invention. FIG. 2 is an example of a first set of CT scans and a second set of CT scans with improved soft tissue contrast, in accordance with some embodiments of the present invention. 1 is a diagram of a system for improving soft tissue segmentation, according to some embodiments of the invention. 2 is an example of a method for improving soft tissue segmentation in accordance with some embodiments of the invention. 1 is a front cross-sectional view of a hybrid computed tomography (CT) scanner according to some embodiments of the invention; FIG. FIG. 4 illustrates an example of a multiscale association linking radiological scale plaque morphology to molecular determinants in accordance with some embodiments of the invention. 1 is an example of multiscale modeling in accordance with some embodiments of the invention. 2 is an example of multiscale modeling enhanced in accordance with some embodiments of the present invention. 3 is an example output screen showing multiple simulated event-free survival probabilities without treatment and under various treatment scenarios, in accordance with some embodiments of the present invention; FIG. 4 illustrates a method for determining and displaying mixed tissue types in accordance with some embodiments of the invention. 2 illustrates a method for determining and displaying mixed tissue types of microcalcification screens and densely calcified areas using radiographic images of multi-energy spectral CT images, in accordance with some embodiments of the present invention. 11 is a diagram illustrating an example of the method of FIG. 10, according to some embodiments of the invention.

It will be understood that, for simplicity and clarity of the drawings, the elements shown in the figures have not necessarily been drawn exactly or to scale. For example, the dimensions of some elements may be exaggerated relative to other elements for clarity, and multiple physical components may be included in a single functional block or element.

In general, the invention provides methods for improving soft tissue contrast, characterizing tissues, classifying phenotypes, stratifying risk, and/or performing multiscale modeling assisted by multi-energy or multi-contrast excitation and evaluation. including systems and methods for performing the same. The present invention can include the application of single-relative multiphase image acquisition, and can also include the use of wide-spectrum CT to assess atherosclerotic plaque tissue in, for example, vessel walls and/or perivascular spaces. It can also be included.

In general, the invention may involve using a software approach to exploit the differential responses of tissues to multi-energy or multi-spectral image sets. The present invention may further include hardware configurations that can further identify a clinically relevant range of tissue. Spectral images acquired over multiple energy levels and/or broad spectra can form the input of one or more algorithms that can improve tissue segmentation.

In some embodiments, the algorithm takes into account that the noise is similar across multiple energies, but the tissues are different, and eliminates the nonlinearity for different tissues by applying an averaging technique that results in a higher signal-to-noise ratio. It may also include using a reaction.

In some embodiments, different tissues can be resolved better at one energy level than at another energy level. In some embodiments, different energies are selected for each tissue. In some embodiments, a digital subtraction approach is used to exploit non-linear responses to different tissues. In some embodiments, different tissue responses from each of the optimal energy levels are combined and dependencies between classes are removed through a multivariate statistical approach. In various embodiments, each of these aforementioned embodiments uses hardware that amplifies such differences, whether designed based on material composition or other physical phenomena, to achieve benefits. Applies to configuration. Examples include tissues distinguished by molecular properties, cellular and molecular environment, material density distribution, morphological expression, response to stimuli, and the like.

In some embodiments, dispersed tissue types are used as an overlay of locally organized tissue.

FIG. 1 is an exemplary radiation input 110 and processed output diagram 100 according to some embodiments of the invention. The processed output is an image 120 that identifies three-dimensional regions for various measures of calcification, lipid-rich necrotic core (LRNC), and matrix, as well as tables and graphs of data related to the various measures. The image may include an image 130 exhibiting a . As will be appreciated by those skilled in the art, the measurands shown in FIG. 1 are examples and may include other measurands.

FIG. 2 is a diagram 200 illustrating an example of when atherosclerosis progresses with common drivers across different degrees of arterial beds based on evolutionary differences and local factors, in accordance with some embodiments of the present invention. It is.

Characterization of coronary artery disease can be accomplished by evaluation of two related but distinct mechanisms. Two are the ability of the arterial system to meet the demands under stress that require delivery of more oxygenated blood to downstream perfused tissues, i.e., stress-induced ischemia (e.g., as FFR or iFR). ), and the tendency of the plaque to physically rupture or form emboli, ie, infarction risk (HRP). Optimally select a treatment pathway based on the processed outputs (e.g., the processed outputs described above in Figure 1 and below) to determine the extent to which each of the different mechanisms is present, e.g., one It can be characterized as being present but not the other, neither being present, or both being present. In some embodiments, the low or high probability of a condition is used to predict the time to occurrence of an adverse event at the individual patient level. Recommendations can be made based on the FRR/HRP ratio. Recommendations can be made based on the likelihood of the condition and may include maintenance medications, intensive drug therapy, symptomatic treatment, or revascularization.

In some scenarios, it is desirable to identify tissue types (eg, complex tissues) that have complex cellular or molecular environments that manifest as overlapping material densities. Current methods can be incorrect and may not correctly identify tissue types. Current methods may involve simple thresholding of material density given by Hunsfield units (HU) of voxels. For example, LRNC or idiopathic pulmonary hemosiderosis (IPH) may not be easily and/or reliably identified and/or the biological treatments that cause them and/or the organisms that are induced in response to them. The scientific process can be complex. LRNCs are variably composed of lipid deposits, cholesterol crystals, apoptotic cell debris, macrophages, and/or calcification. IPH is variably composed of intact and/or ruptured red blood cells, macrophages, hemorrhagic debris, fibrin, cholesterol crystals, and/or calcifications. Complex tissues can present difficulties for systems using a single excitation energy in CT or a characteristic RF profile in MR, and the heterogeneity of plaque composition may lead to differences in tissue response to excitation, e.g. Tissues respond differently to various energies based on how they react to one energy versus another or one radio frequency versus another. There is a possibility that it may be shown. Therefore, it is desirable to relax the strict material density dependence. In some embodiments, different density distributions can be mathematically fitted to samples from histopathology to alleviate strict material density dependence.

In some embodiments, multi-energy signals are applied during the CT scan and digital subtraction, averaging, and/or selection techniques are used to amplify the signal of one tissue type over another, can be used to relax the strict material density dependence. In some embodiments, by maximizing the Mumford-Shah function, the HU of neighboring voxels can be analyzed as a distribution and not just as individual voxels, as an exemplary embodiment.

It is understood that similar to the staining applied to tissue that helps pathologists make decisions when outlining LRNC and IPH, the present invention applies "digital staining" to images to achieve similar amplification. These tissues are heterogeneous, and overlapping tissues are composed of multiple cell types with different densities, and each tissue responds differently to different incident energies. be. Pathological annotation with this in mind is similar to what a pathologist does for histology, so that when a pathologist marks the tissue border of LRNC and IPH on CTA, clinical insight based on that is applied. need to be established using judgment. Applying a HU threshold is not sufficient for this. It requires a mathematical formalism that can be used to mimic the judgment used by human pathologists trained to identify tissues.

MRI, ultrasound, nuclear medicine, and/or other imaging modalities can be reconstructed at various strengths and weaknesses prior to post-processing. FIG. 3 is an example of a first set of CT scans 310a, 320a and a second set of CT scans 310b, 320b with improved soft tissue contrast, in accordance with some embodiments of the invention. One CT scan 310a in the first set of CT scans shows two regions 330a, 340a, which are shown in the second set of CT scans 330a, 340a, as seen in 330b, 340b. Improved by one CT scan 310b in the scan. One CT scan 320a in the first set of CT scans shows one region 350a, which is improved in one CT scan 320b in the second set of CT scans, as seen in 350b. be done.

FIG. 4 is a diagram of a system 400 for improving soft tissue segmentation, according to some embodiments of the invention. The system may include an imaging device 410 and a classification unit 430.

Imaging device 410 may be a CT scanner, magnetic resonance imaging device, ultrasound device, and/or other imaging device known in the art. Imaging device 410 can acquire one or more radiographic images (eg, CT, MR, ultrasound) 420. The plurality of radiographic images 420 may be monochromatic. Multiple monochromatic images 420 can be taken with multiple excitations. As one skilled in the art will appreciate, the images taken with multiple excitations may be the result of transmission, reception, or any combination thereof.

The multiple excitations may include multiple energies, radio frequency pulses (eg, sequences), and/or different frequencies or timing, such as those used by CT, MRI, and/or ultrasound, respectively. In some embodiments, each of the plurality of monochromatic images 420 has a unique excitation of the CT scanner (e.g., a unique x-ray energy, transmitted, received, or both, as a narrow energy range or as a wide (as a range), the unique pulse sequence of MRI, the unique frequency and/or timing of ultrasound.

The number of multiple monochromatic images 420 may depend on the type of tissue or phenotype desired to be identified.

In some embodiments, some monochromatic images 420 of the plurality of monochromatic images 420 are taken with the same excitation, and other monochromatic images 420 are taken with different excitations. For example, the excitation may change each time two monochromatic images 420 are taken. In another example, only the first two monochromatic images 420 are of the same excitation, and the remaining monochromatic images 420 are taken with unique excitations. As will be apparent to those skilled in the art, the plurality of monochromatic images 420 may include any combination of the same and unique excitations.

Excitation may depend on the type of tissue expected. For example, for an expected tissue type of calcium and a CT imaging device, the excitation (transmitted, received, or both) may be 4 keV. In another example, for a tissue type with expected oxygen and a CT imaging device, the excitation (transmitted, received, or both) may be 530 eV.

In the example shown in FIG. 4, a plurality of monochromatic images 420 are shown taken with a CT imaging device and an exemplary range of excitations (transmitted, received, or both) from 65 KeV to 130 KeV; The excitation may be, for example, as low as 0.5 keV or as high as 400 keV, depending on the tissue of interest; such a spectrum allows identification of the tissue type identified as 440 in particular.

Classification unit 430 may include multiple processes that analyze multiple monochromatic images 420. The plurality of processes may include a digital subtraction process, a digital addition process, a multivariate statistical process, and/or an excitation selection process, as described in further detail below with respect to FIG. Classification unit 430 may output data 440 indicative of one or more classified tissue types (e.g., segmented image data of tissue types and/or other data shown in FIG. 1 above). can.

The output data 440 may be of tissue type: blood leak, macrophage, angiogenesis, CALC map, LRNC map, FRESH IPH map, MATX map, microcalcification, erosion, OLD IPH map, thrombus, or any combination thereof. Can be classified. As will be apparent to those skilled in the art, the output data may be of tissue types known in the art.

FIG. 5 is an example of a method 500 for improving soft tissue segmentation, according to some embodiments of the invention. The method may include acquiring multiple radiographic images (eg, CT images, MRI images, ultrasound images). Radiographic images can be acquired by a hybrid CT imaging device shown in FIG. 6 below. Each of the plurality of radiographic images may be acquired using a different excitation (step 510).

In some embodiments, the plurality of radiographic images are CT images, and the different excitations include different x-ray energies (transmitted, received, or both, as a narrow energy range or as a wide range). ). In some embodiments, the plurality of radiographic images are MR images and the different excitations are different radio frequency pulses. In some embodiments, the plurality of radiological images are ultrasound images and the different excitations are at different frequencies.

The method may include selecting one treatment from the plurality of treatments (step 520) to analyze the plurality of radiological images based on the expected soft tissue type (eg, by classification unit 430). The plurality of processes may include digital subtraction processes, digital addition processes, multivariate statistical processes, and/or excitation selection processes.

The selection of one treatment from multiple treatments may be random, based on user input, and/or based on tissue characteristics.

The digital subtraction process may include taking one of the plurality of radiographic images in a certain spectral range and subtracting it from another of the plurality of radiographic images. In some embodiments, digital subtraction may include subtracting a first subset of the plurality of radiological images from one or more of the plurality of radiological images not within the subset. In some embodiments, the digital subtraction process may include subtracting a narrow range from a wide range, for example, to make the narrowband signal higher than the wideband noise floor. In some embodiments, the digital subtraction process may include subtracting one intermediate bandwidth from another intermediate bandwidth to distinguish between two related tissue types.

The digital addition process may include averaging multiple radiological images. Averaging multiple radiographic images is generally advantageous because the noise is similar across energy, but for tissue signals it is not advantageous because averaging can result in a high signal-to-noise ratio.

Multivariate statistical processing combines each of multiple radiological images with mathematical operators other than simple subtraction or addition, and a multivariate statistical approach from a set of techniques identified as nonlinear operators to reduce interclass dependence. and removing the relationship.

Because a particular tissue type can be resolved well at one energy level or a subset of energy levels, the excitation selection process may be used to select one or more of the images for a particular tissue type. It may include selecting a subset. For example, the energy dependence of one tissue may be at a stable point at a high or low kVp level, while the energy dependence of another tissue may be in a different range, which may be One range is used and suggests other ranges are used for other organizations.

The processed plurality of radiological images can be input into one or more classification models. In some embodiments, the classification model is trained as described in U.S. Pat. You can.

The method may include segmenting the plurality of processed radiographic images to display soft tissue (step 530). In some embodiments, segmenting the processed plurality of excitations further includes segmenting the medical image data into three-dimensional (3D) objects.

In some embodiments, segmenting the processed plurality of radiological images includes segmenting the processed plurality of radiological images into exterior wall boundaries. In some embodiments, the segmentation includes segmenting the plurality of radiographic images into lumens and outer walls based on segmented luminal boundaries, outer walls, perivascular regions, and/or focal tissue boundaries. include. FIG. 4 of US Pat. No. 11,094,058 shows an example of segmentation levels for a multi-scale vessel wall analyte map, with corresponding description.

FIG. 6 is a front cross-sectional view of a hybrid computed tomography (CT) scanner 600, a processor 670, and two analog-to-digital converters 655, 665, according to some embodiments of the invention. Hybrid CT scanner 600 can provide truncated spectral scans and/or local reconstructions constrained to global grayscale images from conventional CT configurations.

Hybrid CT scanner 600 includes a first X-ray source 610, a second X-ray source 620, a photon counting detector 630, an energy integrating detector 640, and a gantry 650. The energy integrating detector 640 may be arranged at a predetermined angle between the photon counting detector 630 and the second X-ray source 620 with respect to the first X-ray source 610.

As shown in FIG. 6, the predetermined angle is approximately 90 degrees. The predetermined angle can be based on optimizing the physical space around the circular arc so that the integrating detector can cover a large area.

The first X-ray source 610 and/or the second X-ray source 620 may be a polychromatic X-ray source. First x-ray source 610 and/or second x-ray source 620 may be any x-ray source known in the art for use in CT scanners.

In operation, a patient 660 enters the hybrid CT scanner 600 and the first x-ray source 610 can direct a first x-ray attenuation onto the patient 660 and the energy integrating detector 640. Energy integrating detector 640 may be sized to generate an image of a predetermined region of patient 660.

Energy integrating detector 640 has a curvature defined by radius R and diameter D and can extend along arc A such that a predetermined region of the patient is imaged. For example, radius R may be 30 inches, diameter D may be 60 inches, and arc A may be 135 degrees.

In some embodiments, the first x-ray attenuation is transmitted in the range of 65-130 keV or wider.

Energy integrating detector 640 receives at least a portion of the first x-ray attenuation and the energy transmitted through patient 660.

A second x-ray source 620 can direct a second x-ray attenuation to a patient 660 and a photon counting detector 630. Photon counting detector 630 can generate images of specific tissue targets within a predetermined area.

In some embodiments, the second x-ray attenuation is transmitted in a similar range or a different range than the first embodiment.

Photon counting detector 630 receives at least a portion of the second x-ray attenuation and the energy transmitted through patient 660.

Energy integrating detector 640 and photon counting detector 630 each transmit the energy they receive to processor 670 via A/D converters 655 and 665.

It is known that compressive sensing-based reconstruction algorithms can be used to accurately reconstruct regions of interest when objects are piecewise constant. TV minimization-based intraspectral reconstruction finds important applications in the CT field, such as dose reduction, fast data acquisition, easy data storage, and hardware cost reduction. The assumption made in many reconstruction schemes that the imaged object is piecewise constant is due to tissue composition and/or contrast agents having gradient densities that are significantly smaller than the voxel scale. , which is usually not satisfied in clinical CT imaging, so to avoid this assumption, we use global energy integrated images to facilitate intraspectral CT reconstruction and use processor 670 to significantly reduce the data Minimize reconstruction errors due to truncation.

Processor 670 receives output from energy integrating detector 640 and photon counting detector 630 and processes the images to generate images based on images of predetermined regions of the patient and specific tissue targets within the predetermined images. Generate the final CT image. The final CT image can be used as input for the processing described above in FIG.

Processor 670 may include multiple processing modules. The processing module may include a module 671 that receives the spectral CT from the photon counting detector 630 from the A/D converter 655. The spectral CT can be sent to a photon counting detector signal processing module 672 to create a sinogram. The sinogram is sent to local scan module 673 for reconstruction. Module 675 may receive grayscale CT from energy integrating detector 640 from A/D converter 675 . The grayscale CT can be sent to an energy integrating detector processing module 676 to create a sinogram. The sinogram is sent to global scan module 677 for reconstruction. The outputs of local scan module 673 (e.g., narrowband outputs) and the outputs of global scan module 677 (e.g., multi-energy outputs (e.g., multiple narrowband outputs or Single wideband output)).

As is known in the art, the processing modules on processor 670 may be implemented with one processor or multiple processors.

Energy integrating detector 640 may be considered a global scan of patient 660, while photon counting detector 630 may be considered a local scan. The CT images from energy-integrating detector 640 can be used for standalone CT reconstruction or with global grayscale constraints for intraspectral CT local tomography reconstruction using photon-counting detector 630. It can be used as Photon counting detector 630 may include multiple energy bins configured to image a particular tissue target, using multiple channels of the reconstructed global grayscale constrained spectral image to represent the tissue. can be distinguished.

An intraspectral CT reconstruction using grayscale images as a global constraint can be determined as follows.

のように記述することができる。ここで、特定のスペクトルチャネルのμ_ｉ（ｒ）は、

のように、グレースケールテンプレートからの減衰分解として表すことができる。ここで、μ_ｇｒａｙ（ｒ）は、エネルギー積分投影から再構成された減衰マップである。

即ち、

である。 Assume that N is the number of spectral channels (eg, the number of energies or tube voltages in a CT image), I _i is the photon intensity of a particular spectral channel, and i is the channel index. The energy integral projection image is

It can be written as: Here, μ _i (r) of a particular spectral channel is

can be expressed as a decay decomposition from a grayscale template, as in where μ _gray (r) is the attenuation map reconstructed from the energy integral projection.

That is,

It is.

が得られる。 Assuming that δμ _i (r) is sufficiently small using Taylor series expansion and ignoring higher order terms, then

is obtained.

であると仮定し、それを式５に代入して、

が得られる。

Assuming that, and substituting it into equation 5,

is obtained.

である。 In addition,

It is.

が得られる。 Assuming there are three channels, i.e. N=3,

is obtained.

から、以下の線形逆方程式が得られる。

From this, we get the following linear inverse equation:

where N is the total number of spectral channels and I is the photon intensity emitted from the X-ray source. I _i is the photon intensity in the spectral channel, i is the spectral channel index, μ is the reconstructed attenuation map, B is the photon intensity detected from the spectral channel, and G is , is the detected grayscale photon intensity, r is the spatial position in 3D space, δ is the small change, and exp(x) is the exponential function.

FIG. 7A shows an example of a multiscale association linking radiological scale plaque morphology to molecular determinants, according to some embodiments of the invention. For example, prediction-based scoring, including the validation of surrogate markers, can improve (1) dynamic vascular performance (e.g., hyperemia (e.g., causing ischemia) and/or rupture risk (e.g., causing infarction) ( 2) can be used for For example, quantification of morphometric quantities, including but not limited to IPH, includes analysis (3) of modalities including, for example, monoenergetic CTA or multispectral CTA. Fluid mechanics, e.g., shear stress evaluation and finite element modeling (FEM) can elucidate mechanical triggers of smooth muscle cell (SMC) differentiation (4) and demonstrate (2).

FIG. 7B is an example of multi-scale modeling according to some embodiments of the invention. In FIG. 7B, x represents the predisposition and/or expression of species at the cellular and/or molecular level (e.g., to characterize tissues), and y is assayed by radiology and/or to classify the phenotype. represents the presentation of the macromolecular tissue used, and z represents the predicted outcome (such as risk stratification), simulated progression, and/or simulated regression under different classes of treatment. If x and y are known, x and y can be used to train a model, generate a hypothesis, and/or provide a basis for a simulation. If x is known but y is not known, or vice versa, a predictive model can be utilized if trained using examples where both x and y are known. One exemplary objective is to provide a means that is practically and/or economically easy to implement clinically to obtain information across scales.

In some embodiments, causal relationships can be established analytically, such as whether x causes y and/or can a mechanism be identified that if y is seen, then at the level of x, it probably produces y. . In some embodiments, the utility of such an analysis is that by knowing x, treatment can be individualized if one knows what particular drug or surgical intervention may be optimal. including.

In some embodiments, the analysis method spatially differentiates the relationship between x and y to account for the spatial context, e.g., the value of single-cell techniques over techniques that do not provide a distinction. Avoiding consequential dilutions can be used to establish whether specificity can be determined to increase diagnostic reliability.

FIG. 7C is an example of enhanced multi-scale modeling according to some embodiments of the invention. Extending multiscale modeling techniques, x, y, and z may include the addition of time-dependent functions. For example, x(t) can represent the development of plaques due to premature aging and/or accounting for other mechanisms at the cellular/molecular level, and y(t) can represent macroscopic observations observed over time in radiology. , and z(t) can predict what will happen next under a candidate treatment regimen (including, for example, if left untreated). In some embodiments, the analysis provides y(t) as a histologically validated 3D object on a spatially resolved basis at a macroscopic level, and/or in a convolutional neural network for phenotypic classification. appear. To connect to the molecular/cellular level, you can mine the literature and use tissue resources to enhance de novo experiments for data collection.

FIG. 8 is an example output screen 800 showing multiple simulated event-free survivability probabilities without treatment and under various treatment scenarios, according to some embodiments of the invention. Simulated event-free survival probabilities include systemic inflammatory treatment 810, intensive lipid lowering treatment 820, statin s 830, revascularization (eg, stent) 840, and no treatment 850.

Identifying complex tissue presentations (e.g., tissue within a region containing more than one tissue type at a particular location) whether acquiring a single-energy or multi-energy CT scan I can do it. For example, in complex tissue areas, microcalcifications may occur, such as in MATX or LRNC. Adaptive grid sizing and/or adaptive region growth can be used to create subtle representations of complex tissue types. Complex biology can be represented as a screen overlay or grid point overlay, and the progression of complex biology as well as the final result can be determined. For example, the final outcome of dense macrocalcifications as well as the progression of microcalcifications can be determined.

Such a screen or grid point overlay can typically enhance the presentation of focal tissue without replacing it. Continuing with the example of microcalcifications, in subjective terms the term "patchy calcifications" is used, but this term cannot express intermediate signs of local organization. In some embodiments, intermediate signs of local organization of microcalcifications include a "grid overlay" representing sparse or dense points (e.g., point density rather than material density) to indicate patterns of organization. It can be visually expressed as

Grid overlays can be expressed as "textures" for human observers to visualize in software, and/or as "mottling" on data objects that are well suited for computer processing. I can do it. Thus, tissue organization may affect not only MATX but also LRNC and/or other tissue regions. Similar to IPH, microcalcifications can be subdivided into two tissue/stage types: compact stage and/or early stage. Microcalcifications may become grids, whereas dense macrocalcifications may become focal points. The microcalcification grid generalizes how tissue is represented and can be interpreted as a distribution of foci that can represent a transition state where foci gather. For example, we have not yet reached a congested state, but we are approaching a congested state. Tissue types such as blood leakage, microcalcifications, and angiogenesis can be represented as grids of varying point densities. Tissue types expressed as local regions such as LRNC, CALC, IPH, etc. may become local regions. MATX can be described as a local region, but preferably has regions described as other tissues.

FIG. 9 illustrates a method 900 for determining and displaying mixed tissue types, according to some embodiments of the invention.

The method may include receiving a radiographic image of the patient (step 910). The radiographic image may be a CT image, MR image, or ultrasound image. If the radiographic image is a CT image, a hybrid CT imaging device can be used to acquire the CT image, as illustrated in FIG.

The method includes the step of determining a first tissue type in a region of interest based on a radiographic image of a patient, the first tissue type being a grid of points in the region of interest (step 920). May include. For example, the first tissue type may be any tissue type that can progress from less dense to more dense. The first tissue type may be any tissue type that is distributable to another tissue type. The first tissue type may be blood leakage, microcalcification and/or neovascularization.

Each grid of points can be represented as a point density. Each point on the grid of points may have some, all, or unique density values.

The first tissue type may be determined by the system shown in FIG. 4 above, the method shown in FIG. 5 above, the system shown in FIG. 6 above, or any combination thereof.

The method includes the steps of determining a second tissue type within a region of interest based on a radiographic image of a patient, the second tissue type being a local region within the region of interest; may include the step of at least some grid points of the second tissue type being aligned with grid points of the second tissue type (step 930). The second tissue type may be any tissue type that can be represented as a localized region. The second tissue type may be LRNC, CALC, and/or IPH.

The second tissue type may be determined by the system shown in FIG. 4 above, the method shown in FIG. 5 above, the system shown in FIG. 6 above, or any combination thereof.

FIG. 10 shows a method 100 for determining and displaying mixed tissue types of microcalcification screens and densely calcified areas using radiographic images of multi-energy spectral CT images, according to some embodiments of the present invention. shows.

The method includes receiving (step 1010) a multi-energy spectral CT image (eg, by the system shown in FIG. 4 above).

The method may include performing (1020) multi-energy photon counting K-edge differential imaging. In some embodiments, for K-edge resolved imaging of multi-energy systems with photon-counting detectors (PCDs), the energy bins have a large effect on the intensity of the extracted K-edge signal and a small Provides optimized energy bins with the potential to improve classification between calcification and focal/dense calcification.

The method may include performing denoising of the spectral image using the regularized model (1030). In some embodiments, the linear attenuation coefficient map is decomposed into base materials that are separable in the spectral and spatial domains. The nonlinearity can be translated into a reconstruction of the mass density map. By reducing the dimensionality of the optimization variables and/or by using a minimization scheme that solves the reconstruction with a weighted kernel norm and regularization of the total variation, less noise and more spectral information can be obtained. can be provided.

The method may include, for example, improving (eg, removing) calcium by subtracting A from B (step 1040) to effectively improve the signal-to-noise ratio.

The method may include segmenting the improved calcium (step 1050). Segmentation results in focal areas of dense calcification and microcalcification screens (such as grid overlays).

In some embodiments, principal component analysis (PCA) for feature extraction is used with spectral imaging to extract small HU differences in soft tissue that are present in multi-energy images. Soft tissue can be extracted using, for example, adaptive region growing and K-means clustering techniques. The region growing algorithm can be used to separate regions with the same properties as predefined properties, giving good segmentation results to the original image with sharp edges. In some embodiments, the segmentation step includes: 1) applying morphological reconstruction to smooth flat regions of the image and/or preserve edges; 2) thickening and/or merging edges. 3) use top/bottom hat transformations for contrast enhancement; 4) local to the location of both internal and external markers. 5) using a weighted function to combine the top/bottom hat transformation algorithm and the marker algorithm to obtain a new algorithm; Good too. In this way, over-segmentation from traditional watersheds can be prevented.

FIG. 11 is a diagram illustrating an example of the method of FIG. 10, according to some embodiments of the invention. Diagram 1110 shows a received multi-energy spectral CT image (eg, step 1010 shown in FIG. 10 above).

Diagram 1130 shows the output of performing multi-energy photon counting K-edge differential imaging (eg, step 1020 shown in FIG. 10 above) in a CT image. Diagram 1120 shows the output of performing spectral image denoising using a regularization model on a CT image (eg, step 1030 shown in FIG. 10 above). Diagram 1140 shows the output with calcium removed (eg, step 1040 shown in FIG. 10 above). Diagram 1150 and diagram 1160 illustrate segmenting the removed calcium into dense calcifications 1160 and microcalcifications 1150 (eg, step 1050 shown in FIG. 10 above).

Embodiments of the invention are discussed using terms such as, but not limited to, "processing," "computing," "calculating," "determining," "establishing," "analyzing," and "verifying." Data represented as physical (e.g., electronic) quantities in a computer's registers and/or memory, or instructions for performing operations and/or processes, may be stored in a computer's registers and/or memory. operation of a computer, computing platform, computing system or other electronic computing device to manipulate and/or convert other data similarly represented as physical quantities in other non-transitory information storage media capable of It may also refer to processing.

Embodiments of the invention are not limited to, but the terms "plurality" and "plurality" as used herein may include, for example, "a number" or "two or more." The terms "plurality" or "plurality" may be used throughout this specification to describe more than one component, device, element, unit, parameter, etc. A term set as used herein may include one or more items. Unless explicitly stated, the method embodiments described herein are not limited to any particular order or sequence. Furthermore, some of the described method embodiments or elements thereof may occur or be performed together at the same time, at the same point in time.

A computer program may be written in any form of programming language, including compiled or interpreted languages, and may be a stand-alone program or a combination of subroutines, components, and/or other units suitable for use in a computing environment. It may be developed in any form including. A computer program may be deployed to run on one computer or multiple computers at one site.

The method steps may be performed by one or more programmable processors executing a computer program to perform the functions of the invention by manipulating input data and generating output. The method steps may be performed by an apparatus or implemented as dedicated logic circuitry. The circuit may be, for example, an FPGA (field programmable gate array) and/or an ASIC (application specific integrated circuit). Modules, subroutines, and software agents may refer to portions of computer programs, processors, special circuits, software, and/or hardware that implement the functionality.

Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, as well as any one or more processors of any type of digital computer. Generally, a processor receives instructions and data from read-only memory and/or random access memory. The key components of a computer are a processor that executes instructions and one or more memory devices that store instructions and data. Generally, a computer is operably coupled to receive data from or transmit data to one or more mass storage devices (e.g., magnetic disks, magneto-optical disks, or optical disks) that store data. be done.

Data transmissions and instructions may also occur via a communications network. Information carriers suitable for embodying computer program instructions and data include, by way of example, all forms of non-volatile memory including semiconductor memory devices. The information carrier may be, for example, an EPROM, an EEPROM, a flash memory device, a magnetic disk, an internal hard disk, a removable disk, a magneto-optical disk, a CD-ROM and/or a DVD-ROM disk. The processor and memory may be supplemented by and/or incorporated into dedicated logic circuitry.

To provide user interaction, the techniques described above may be implemented on a computer having a display, a transmitter, and/or a computing device. The display device may be, for example, a cathode ray tube (CRT) and/or a liquid crystal display (LCD). User interaction may be, for example, displaying information to the user, allowing the user to provide input to the computer (e.g., interacting with user interface elements), using a keyboard and pointing device (e.g., mouse or trackball), etc. good. Other types of devices may also be used to provide user interaction. Other devices may, for example, provide feedback to the user in any form of sensory feedback (eg, visual, auditory, or tactile feedback). Input from the user can be received in any form, including, for example, acoustic input, voice input, and/or tactile input.

A computing device may be, for example, a computer, a computer with a browser device, a telephone, an IP phone, a mobile device (e.g., a cell phone, a personal digital assistant (PDA) device, a laptop computer, an e-mail device), and/or another computer. It may also include a communication device. A computing device may be, for example, one or more computer servers. A computer server may be part of a server farm, for example. The browser device may include, for example, a World Wide Web browser (e.g., Microsoft® Internet Explorer® available from Microsoft Corporation, Chrome available from Google, Mozilla® Firefox available from Mozilla Corporation). , Safari available from Apple) (e.g., desktop computers, laptop computers, tablets). Mobile computing devices include, for example, personal digital assistants (PDAs).

Websites and/or web pages can be provided, for example, over a network (eg, the Internet) using a web server. The web server may include, for example, a server module (e.g., Microsoft® Internet Information Services available from Microsoft Corporation, Apache Web Server available from the Apache Software Foundation, Apache Tomcat Web Server available from the Apache Software Foundation). ) It may be a computer equipped with

Storage modules include, for example, random access memory (RAM) modules, read-only memory (ROM) modules, computer hard drives, memory cards (e.g., universal serial bus (USB) flash drives, secure digital (SD) flash cards), floppy It may be a disk and/or any other data storage device. Information stored in the storage module may be maintained, for example, in a database (eg, a relational database system, a flat database system) and/or any other logical information storage mechanism.

The techniques described above can be implemented in a distributed computing system that includes back-end components. A backend component may be, for example, a data server, a middleware component and/or an application server. The techniques described above can be implemented in a distributed computing system that includes a front-end component. The front end component may be, for example, a client computer having a graphical user interface, a web browser that allows a user to interact with the implementation, and/or other graphical user interface for the sending device. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), the Internet, wired networks, and/or wireless networks.

The system may include a client and a server. Clients and servers are generally remote from each other and typically interact via a communications network. The relationship between clients and servers occurs through computer programs running on their respective computers and having a client/server relationship with each other.

The networks described above may be implemented as packet-based networks, circuit-based networks, and/or a combination of packet-based networks and circuit-based networks. Packet-based networks include, for example, the Internet, carrier Internet Protocol (IP) networks (e.g., local area networks (LANs), wide area networks (WANs), campus area networks (CANs), metropolitan area networks (MANs), home area network (HAN), private IP network, IP private branch exchange (IPBX), wireless network (e.g., radio access network (RAN), 802.11 network, 802.16 network, general packet radio service (GPRS) network, high performance wireless LAN), and/or other packet-based networks. Circuit-based networks may include, for example, the public switched telephone network (PSTN), private branch exchange (PBX), wireless networks (e.g., RAN, Bluetooth® ), code division multiple access (CDMA) networks, time division multiple access (TDMA) networks, Global System for Mobile Communications (GSM) networks), and/or other circuit-based networks.

Some embodiments of the invention may be embodied in a system, method, or computer program product. Similarly, some embodiments may be implemented in hardware, software, or a combination of both. Some embodiments may be embodied as a computer program product stored in the form of computer readable program code on one or more non-transitory computer readable media. Such non-transitory computer-readable medium may include instructions that, when executed, cause a processor to perform method steps in accordance with embodiments. In some embodiments, instructions stored on a computer-readable medium may be in the form of installed applications and installation packages.

Such instructions may be loaded and executed by one or more processors, for example. For example, a computer-readable medium may be a non-transitory computer-readable storage medium. The non-transitory computer-readable storage medium may be, for example, an electronic, optical, magnetic, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof.

Computer program code may be written in any suitable programming language. The program code may be executed by a single computer system or by multiple computer systems.

Those skilled in the art will appreciate that the invention may be embodied in other specific forms without departing from its spirit or essential characteristics. Accordingly, the embodiments described above are to be considered in all respects as illustrative rather than limiting on the invention described herein. The scope of the invention is, therefore, indicated by the appended claims rather than by the foregoing description, and all changes that come within the meaning and range of equivalency of the claims are intended to be embraced therein.

In the above detailed description, many specific details are set forth to provide an understanding of the invention. However, one of ordinary skill in the art will understand that the invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, modules, units and/or circuits may not be described in detail so as not to obscure the present invention. Certain features or elements described with respect to one embodiment may be combined with features or elements described with respect to other embodiments.

Claims (24)

A computerized method for improving soft tissue analysis, comprising:
acquiring a plurality of radiographic images of a patient by a computing device, each of the plurality of radiographic images being acquired using a different excitation;
selecting, by the computing device, one treatment from the plurality of treatments to analyze the plurality of excitations based on the expected soft tissue type;
segmenting, by the computing device, the processed plurality of excitations to display the soft tissue;
computerized methods, including; 2. The computerized method of claim 1, wherein the plurality of radiological images are computed tomography (CT) images and the different excitations are different x-ray energies. 2. The computerized method of claim 1, wherein the plurality of radiographic images are magnetic resonance (MR) images and the different excitations are different radio frequency pulses. 2. The computerized method of claim 1, wherein the plurality of radiological images are ultrasound images and the different excitations are at different frequencies. determining, by the computing device, a first tissue type within a region of interest based on the plurality of radiographic images of the patient, wherein the first tissue type is defined by a grid of points across the region of interest; The steps represented and
determining, by the computing device, a second tissue type within the region of interest based on the plurality of radiographic images of the patient, wherein the second tissue type is locally within the region of interest; a region, wherein at least some grid points of the first tissue type are coincident in position with grid points of the second tissue type;
2. The computerized method of claim 1, further comprising: 2. The computerized method of claim 1, wherein the plurality of processes includes a digital subtraction process, a digital addition process, a multivariate statistical process, or an excitation selection process. 7. The computerized method of claim 6, wherein the digital subtraction process includes subtracting a first subset of the plurality of radiological images from one or more of the plurality of radiological images not within the subset. method. 7. The computerized method of claim 6, wherein the digital summation process includes averaging the plurality of received radiographic images. The multivariate statistical processing is
combining the plurality of radiographic images;
Removing interclass dependencies using a multivariate statistical approach;
7. The computerized method of claim 6, comprising: The plurality of radiation images are CT images, and each of the plurality of CT images is
directing a first x-ray attenuation by a first x-ray source to an energy-integrating detector dimensioned to produce an image of a predetermined region of the patient;
directing a second x-ray attenuation by a second x-ray source to a photon-counting detector that generates an image of a specific tissue target within the predetermined region;
2. The computer of claim 1, wherein the computer is generated by a processor generating a final CT image based on the image of the predetermined region of the patient and the image of the particular tissue target within the predetermined image. ized method. 7. The computerized method of claim 6, wherein the excitation selection process includes selecting a particular radiographic image from the plurality of radiographic images based on the tissue type. A hybrid computed tomography (CT) scanner, comprising:
a first x-ray source that directs the first x-ray attenuation to an energy-integrating detector dimensioned to generate an image of a predetermined region of the patient;
a second x-ray source that directs a second x-ray attenuation to a photon-counting detector that generates an image of a specific tissue target within the predetermined region;
a processor that generates a final CT image based on the image of the predetermined region of the patient and the image of the particular tissue target within the predetermined image;
A hybrid CT scanner, including: 13. The hybrid CT scanner of claim 12, wherein the energy integrating detector and the photon counting detector are arranged to acquire information from the same field of view. 13. The hybrid CT scanner of claim 12, wherein the image of the predetermined region of the patient is a grayscale CT image. 13. The hybrid CT scanner of claim 12, wherein the image of the particular tissue target within the predetermined image is a spectral CT image. 13. The energy integrating detector is arranged with a 90 degree offset between the photon counting detector and the second X-ray source with respect to the first X-ray source. The hybrid CT scanner described. The processor is configured to analyze a tissue type, the analysis comprising:
selecting one process from a plurality of processes to analyze the final CT image based on expected soft tissue types;
segmenting the final processed CT image to display the soft tissue;
The hybrid CT scanner according to claim 12, comprising: 13. The hybrid CT scanner of claim 12, wherein the photon counting detector includes multiple energy bins configured to image the particular tissue target. A computerized method for determining and displaying mixed tissue types, the method comprising:
receiving, by the computing device, a radiological image of the patient;
determining, by the computing device, a first tissue type within a region of interest based on the radiographic image of the patient, the first tissue type being represented by a grid of points across the region of interest; step,
determining, by the computing device, a second tissue type within the region of interest based on the radiographic image of the patient, the second tissue type being a local region within the region of interest; at least some of the grid points of the first tissue type are coincident in position with grid points of the second tissue type;
computerized methods, including; 20. The computerized method of claim 19, wherein the plurality of radiological images are computed tomography (CT) images, magnetic resonance (MR) images, or ultrasound images. 20. The computerized method of claim 19, wherein the first tissue type and the second tissue type are overlaid. 20. The computerized method of claim 19, wherein the grid of points has varying densities. 20. The computerized method of claim 19, wherein the first tissue type is microcalcified tissue and the second tissue type is LNRC, dense calcification or IPH. performing multi-energy photon counting K-edge differential imaging on the CT image by the computing device;
performing spectral image denoising on the CT image by the computing device using a regularization model;
improving the signal-to-noise ratio of calcium on the K-edge subtracted and denoised CT image by the computing device;
segmenting, by the computing device, the improved calcium image to represent one or both of localized dense calcifications and dispersed microcalcifications;
20. The computerized method of claim 19, further comprising:

Images